Behavioral/Cognitive

Thy1-Expressing Neurons in the Basolateral Amygdala MayMediate Fear Inhibition

Aaron M. Jasnow,1 David E. Ehrlich,2 Dennis C. Choi,2 Joanna Dabrowska,2 Mallory E. Bowers,2

Kenneth M. McCullough,2 Donald G. Rainnie,2 and Kerry J. Ressler2,3

1Department of Psychology, Kent State University, Kent, Ohio 44242, 2Department of Psychiatry and Behavioral Sciences, Behavioral Neuroscience, YerkesResearch Center, Emory University School of Medicine, Atlanta, Georgia 30329, and 3Howard Hughes Medical Institute, Bethesda, Maryland 20814

Research has identified distinct neuronal circuits within the basolateral amygdala (BLA) that differentially mediate fear expressionversus inhibition; however, molecular markers of these populations remain unknown. Here we examine whether optogenetic activationof a cellular subpopulation, which may correlate with the physiologically identified extinction neurons in the BLA, would differentiallysupport fear conditioning versus fear inhibition/extinction. We first molecularly characterized Thy1-channelrhodopsin-2 (Thy1-ChR2-EYFP)-expressing neurons as a subpopulation of glutamatergic pyramidal neurons within the BLA. Optogenetic stimulation of theseneurons inhibited a subpopulation of medial central amygdala neurons and shunted excitation from the lateral amygdala. Brief activa-tion of these neurons during fear training disrupted later fear memory in male mice. Optogenetic activation during unreinforced stimulusexposure enhanced extinction retention, but had no effect on fear expression, locomotion, or open-field behavior. Together, these datasuggest that the Thy1-expressing subpopulation of BLA pyramidal neurons provide an important molecular and pharmacological targetfor inhibiting fear and enhancing extinction and for furthering our understanding of the molecular mechanisms of fear processing.

IntroductionIn humans, excessive associative fear is the hallmark of a numberof neuropsychiatric disorders, including specific phobias, socialphobias, panic disorder, and posttraumatic stress disorder(American Psychiatric Association, 2000; Kessler et al., 2005).Classical fear conditioning is perhaps the most powerful experi-mental model for studying the neural basis of associative learningand memory formation in mammals (Maren and Fanselow,1996; Davis, 2000; LeDoux, 2000). In classical fear conditioning,subjects are presented with a neutral conditioned stimulus (CS),usually a tone, which is paired with an aversive unconditionedstimulus (US), usually a foot shock. After successive pairings, thetone acquires aversive properties and elicits fear responses whenpresented alone. In rodents, this fear is frequently quantified bymeasuring freezing behavior displayed during the presentation ofthe tone. Subsequently, when the tone is presented repeatedly inthe absence of the shock, fear responses decline, a phenomenoncalled extinction. Extinction is not the forgetting of the originalassociative memory, but is rather thought to involve new learn-ing, forming a new memory representation that the CS no longerpredicts the US (Bouton et al., 2006; Myers and Davis, 2006).Considerable evidence using a variety of experimental models

indicates that the amygdala is essential in the acquisition, consol-idation, and extinction of conditioned fear (Davis, 2000; Myersand Davis, 2006; Herry et al., 2010).

A major challenge to understanding how the amygdala regu-lates emotional memories is identifying the neuronal circuits thatdifferentially regulate fear acquisition, consolidation, and extinc-tion. Recently, two electrophysiologically distinct populations ofneurons within the basolateral amygdala (BLA) were identified toregulate fear responses in awake-behaving mice (Herry et al.,2008). Specifically, one population of neurons responded to theCS with increased firing frequency during and after fear condition-ing. These same neurons reduced firing frequency during extinctiontraining. A second population only responded to the CS with in-creased firing during extinction training. Importantly, the activity ofthese distinct neuronal populations predicts the behavioral output;increased firing of “extinction” neurons and decreased firing of“fear” neurons preceded a decrease in freezing behavior. These datasuggest that there are distinct subpopulations of neurons within theBLA that are responsible for fear activation and fear inhibition. Todate, however, the molecular identities of these neuronal popula-tions have not been determined. The present study elucidates therole of a specific subpopulation of BLA neurons that may play acritical role in fear processing.

Here, we use a combination of optogenetic manipulation of neu-ronal activity in awake behaving mice, slice electrophysiology, im-munocytochemistry, and in situ hybridization to identify specificneurons within the amygdala regulating long-term fear memories.We used two strains of mice with transgene expression driven by theThy1-promoter (Thy1-YFP; Thy1-ChR2-EYFP), which target a spe-cific subpopulation of pyramidal neurons within amygdala andother forebrain areas (Sugino et al., 2006). This enabled us to exam-ine and manipulate neuronal activity within the same subpopulation

Received Nov. 30, 2012; revised May 6, 2013; accepted May 12, 2013.Author contributions: A.M.J. and K.J.R. designed research; A.M.J., D.E.E., D.C.C., J.D., M.E.B., and K.M.M. per-

formed research; A.M.J. and D.G.R. contributed unpublished reagents/analytic tools; A.M.J., D.E.E., and D.C.C. ana-lyzed data; A.M.J. and K.J.R. wrote the paper.

This work was supported by The Center for Behavioral Neuroscience of the National Science Foundation underAgreement No. IBN-9876754, NIH Grant DA019624, and the Yerkes National Primate Research Center Base Grant2P51RR000165-51.

Correspondence should be addressed to Dr. Kerry Ressler, Howard Hughes Medical Institute, Behavioral Neuro-science, 954 Gatewood Drive NE, Atlanta, GA 30329. E-mail: kressle@emory.edu.

DOI:10.1523/JNEUROSCI.5539-12.2013Copyright © 2013 the authors 0270-6474/13/3310396-09$15.00/0

10396 • The Journal of Neuroscience, June 19, 2013 • 33(25):10396 –10404

of Thy1-labeled glutamatergic neurons within the BLA during andfollowing fear conditioning and fear extinction learning.

Materials and MethodsAnimalsAdult (6 –10 weeks) B6.Cg-Tg(Thy1-COP4/EYFP)18Gfng/J (Thy1-ChR2-EYFP), and B6.Cg-Tg(Thy1-YFP)HJrs/J (Thy1-YFP) mice werepurchased from Jackson Laboratories. All experimental mice weregroup housed (4 –5 mice per cage) on a 12:12 h light:dark cycle in atemperature-controlled colony room and had unrestricted access to foodand water. All procedures were conducted in accordance with policyguidelines set by the National Institutes of Health and were approved bythe Emory University Institutional Animal Care and Use Committee.

Surgical proceduresFor guide cannula implantation, mice were anesthetized deeply with aketamine/dormitor mixture and placed in a stereotaxic instrument. Ste-reotaxic coordinates were derived from Paxinos and Franklin (2004).The head was positioned in the stereotaxic instrument so that the skullwas level between lambda and bregma before implantation of the guidecannulas. Mice were implanted with a unilateral cannula aimed just dor-sal to the BLA. Coordinates were 1.5 mm posterior, 3.4 mm lateral, and4.8 mm below bregma. The side of guide cannula implantation was var-ied across mice. Dummy stylets were placed in the guide cannulas to keepthem patent. Following surgery, mice were handled daily and theirdummy stylets removed and replaced to habituate the mice to beingrestrained and to the insertion of the fiber optic.

Behavior and optogenetic stimulationCue-dependent fear conditioning. Mice were pre-exposed to the condi-tioning chambers (Med Associates) 2 d before training. During cued-feartraining, mice received five paired CS tones (30 s, 6 kHz, 75– 80 db) andUS shock (1 s, 0.5 mA) trials with a 90 s intertrial interval (ITI).

Cued-fear expression. The expression of fear memory was tested 24 hafter fear conditioning in a novel context. Mice were presented with 1530 s tones with a 60 s ITI. Freezing during tone presentation was mea-sured with FreezeView software (Coulbourn Instruments).

Fear extinction, retention, renewal, and reinstatement. Fear extinction wasmeasured 24 h after fear conditioning in a novel context. We used a subop-timal extinction procedure in which mice were presented with 15 30 s toneswith a 60 s ITI. Freezing during tone presentation was measured with Freez-eView software (Coulbourn Instruments). Forty-eight hours after initial ex-tinction training, mice were tested for extinction retention as above. To testrenewal, mice were placed back in the training context and presented withfive initial training CS tones (30 s, 6 kHz, 75–80 db). To test fear reinstate-ment, mice were placed in an entirely new context and presented with asingle unsignaled footshock. Mice were tested 24 h later and presented withfive initial training CS tones (30 s, 6 kHz, 75–80 db).

Open field. Open-field tests were conducted in a dimly lit room. Micewere implanted with fiber optics, then placed in the open-field apparatus(Med Associates) and allowed to explore for 5 min. Following the initialexploration time, mice were allowed to explore for an addition 5 minwith optogenetic stimulation.

Optogenetic stimulation. Optogenetic stimulation was performed us-ing a 50 mW, 473 nm fiber-optic-coupled laser (IkeCool). The laser-coupled fiber was connected to the fiber optic that was inserted into theguide cannula with a fiber optic rotary joint (Doric Lenses). Before eachexperiment, mice were restrained and the fiber optic was inserted directlythrough the guide cannula and secured using a modified dust cap (Plas-tics One). Light stimulation parameters were 2 s stimulation at 20 Hz, 15ms pulses. Light power density during stimulation was estimated at14.32–15.91 mW/mm 2. For stimulation during fear acquisition, the lightstimulation coterminated with the tone and shock. For stimulation dur-ing fear extinction, the light stimulation coterminated with each tone.Stimulation parameters during the open-field experiments were thesame as described for fear conditioning and fear extinction except thatpulses were delivered once every minute for the 5 min exploration time.

Dual fluorescent in situ hybridizationThy1-ChR2-EYFP mice were anesthetized and decapitated. Brains wererapidly removed, frozen on dry ice, and stored at �80°C until processing.Tissue was sectioned at 16 �m on a cryostat and mounted on SuperfrostPlus slides (Fisher Scientific). cDNA clones containing the coding se-quence of mouse Vglut1 and the coding sequence for ChR2 were linear-ized with appropriate restriction enzymes. Fluorescently labeledriboprobes were generated using T7 (Vglut1) and SP6 (ChR2) RNA poly-merase (Maxiscript SP6/T7 kit). The Vglut1 riboprobe was labeled withfluorescein and the ChR2 riboprobe was labeled with digoxigenin. Fol-lowing a prehybridization procedure, the sections were hybridized withboth riboprobes at 55°C for 16 h and then subjected to a series of strin-gent washes. Sections were then incubated with anti-fluorescein-polymerized horse-radish peroxidase (POD) and Fab fragments,followed by fluorescent amplification and peroxidase quenching, andthen with anti-digoxigenin-POD, Fab fragments (Roche). Signals wereamplified with the TSA Plus Fluorescein Fluorescence System or TSAPlus Cy5 Fluorescence System (PerkinElmer) following each series ofprimary antibodies. Sections were then stained with Hoechst, washed,and coverslipped with Mowiol mounting medium.

ImmunohistochemistryStandard chromogenic immunohistochemistry was used to determinec-fos expression after optical stimulation. Briefly, free-floating sectionscontaining the BLA were rinsed in PBS, then rinsed in 0.014% phenyl-hydrazine, permeabilized with 0.5% Triton-X in PBS, and incubated for48 h at 4°C with primary antibody against c-fos (1:1000, rabbit anti-c-fos;Santa Cruz Biotechnology). Sections were then rinsed in PBS and incu-bated at room temperature with goat anti-rabbit biotinylated secondaryantibody (1:500) followed by avidin-biotin-peroxidase complex reagent(Vector Laboratories). Sections were rinsed in PBS and staining wasvisualized using a nickel-enhanced DAB reaction. Slides were cover-slipped with DPX and allowed to dry before imaging. C-fos immuno-staining targeting the BLA was performed �1 h after laser lightstimulation of Thy1-ChR2-EYFP mice following the behavioral studiesdescribed below, using similar stimulation paradigms. We found thatwithout stimulation (no stim), there was minimal c-fos activation,whereas with stimulation (stim), there was increased c-fos staining inBLA and lateral central amygdala (CeL) (see Fig. 2C).

Immunofluorescence was used to determine the relative colocalization ofyellow fluorescent protein (YFP) with specific markers for the major sub-populations of BLA interneurons, namely parvalbumin and calbindin. Todetermine relative expression of YFP within the population of BLA principalneurons, we performed additional experiments using an antibody raisedagainst calcium/calmodulin-dependent protein kinase II � (CaMKII�).CaMKII� is exclusively expressed in the BLA principal neurons and not inGABAergic interneurons (McDonald et al., 2002). Free-floating Thy1-YFPmouse brain sections containing BLA were rinsed in PBS, permeabilizedwith 0.5% Triton-X in PBS, and incubated for 48 h at 4°C with the followingprimary antibodies in 0.5% Triton-X solution: polyclonal rabbit anti-parvalbumin antibody (1:1000; Swant), mouse anti-calbindin antibody(1:1000; Sigma Aldrich), and mouse anti-CaMKII� antibody (1:1000;Cell Signaling Solutions). Subsequently, sections were rinsed in PBSand incubated at room temperature for 2 h with either Alexa-Fluor 488goat anti-mouse IgG or Alexa-Fluor 568 goat anti-rabbit IgG (1:500;Invitrogen), depending on the primary antibody host. Sections were thenrinsed, mounted on gelatin-coated glass slides, and coverslipped usingMowiol mounting medium (Sigma Aldrich). Confocal laser scanningmicroscopy was used to obtain high-resolution photomicrographs usingan Orca R2 cooled CCD camera (Hammamatsu) mounted on a LeicaDM5500B microscope (Leica Mircosystems).

Slice electrophysiologyTissue preparation. Acute, coronal slices (300 �m) containing medialcentral amygdala (CeM), BLA, and lateral amygdala (LA) were collectedfrom 2- to 4-month-old Thy1-ChR2-EYFP mice for ex vivo electrophys-iological recordings. Briefly, animals were decapitated under isofluraneanesthesia (Abbott Laboratories) and the brains were removed and im-mersed in ice-cold, oxygenated cutting solution that contained the fol-

Jasnow et al. • Cell-Type Specific Inhibition of Fear J. Neurosci., June 19, 2013 • 33(25):10396 –10404 • 10397

lowing (in mM): 130 NaCl, 3.5 KCl, 1.1 KH2PO4, 6.0 MgCl2, 1.0 CaCl2, 30NaHCO3, 10 glucose, 0.8 thiourea, 2 sodium pyruvate, 0.4 ascorbic acid,and 2 kynurenic acid. A block of tissue containing the amygdala was thenmounted in a Leica VTS-1000 vibrating microtome (Leica Microsys-tems) and coronal slices were cut and hemisected. Slices were transferredto a holding chamber containing oxygenated cutting solution at 32°C for40 min before being placed in room-temperature, oxygenated, artificialCSF (ACSF) containing the following (in mM): 130 NaCl, 3.5 KCl, 1.1KH2PO4, 1.3 MgCl2, 2.5 CaCl2, 30 NaHCO3, 10 glucose, 0.8 thiourea, 2sodium pyruvate, and 0.4 ascorbic acid.

Patch clamp. Slices were placed in a Warner Series 20 recording cham-ber (Warner Instruments) mounted on the fixed stage of a Leica DM-LFSmicroscope (Leica Microsystems). Slices were fully submerged and con-tinuously perfused at a rate of 1–2 ml/min with heated (32°C) and oxy-genated ACSF. Before patch clamping, a concentric bipolar stimulatingelectrode (FHC) was placed in the LA, and a 200 �m fiber-optic cable(Thorlabs) connected to a 488 nm laser (Shanghai Lasers) was placed atthe external capsule, lateral to and aimed at the BLA. The BLA, LA, andCeM were identified using a mouse brain atlas (Paxinos and Franklin,2004), and neurons were selected for recording under IR-DIC illumina-tion with a 40� water-immersion objective. For all experiments, whole-cell patch-clamp configuration was established and cell responses wererecorded in current-clamp mode. For measuring responses to opticalstimulation, pulses of 0.6 –1.2 mW, 488 nm light were delivered at 0.2 or20 Hz. For electrical stimulation, 150 �s pulses were delivered to the LA.Images were captured with a Hamamatsu Orca ER CCD camera con-trolled by SimplePCI software (Compix). Whole-cell patch-clamp re-cordings were conducted using thin-walled borosilicate glass-patchelectrodes (WPI), which were pulled on a P-97 Flaming/Brown micropi-pette puller (Sutter Instruments). Patch electrodes had resistances rang-ing from 4 to 7 M� when filled with standard patch solution thatcontained the following (in mM unless otherwise noted): 138K-gluconate, 2 KCl, 3 MgCl2, 5 phosphocreatine, 2 K-ATP, 0.2 NaGTP,10 HEPES, and 3 mg/ml biocytin. The patch solution was adjusted to apH of 7.3 with KOH and had a final osmolarity of �280 mOsm. Junctionpotentials were offset before patching neurons. Series resistances andinput resistance were monitored online with a 50 pA hyperpolarizingstep (200 ms) given between stimulation sweeps, and neurons with morethan a 15% change in series resistance were discarded. Recordings wereobtained using a MultiClamp 700A amplifier (Molecular Devices), aDigidata 1322A A/D interface, and pClamp 10 software (Molecular De-vices). Data were filtered at 5 kHz and sampled at 10 kHz. Neurons wereexcluded from analysis if their resting membrane potential (Vm) wasmore positive than �55 mV or if their action potentials did not surpass�5 mV. Neurons were also excluded if the recording location was incor-rect based on post hoc visualization of recorded neurons using a LeicaDM5500B confocal, spinning disk laser microscope (Leica Microsys-tems) equipped with an Orca R2 cooled CCD camera (Hamamatsu) afterfixation with 10% buffered formalin (Fisher Scientific) and staining withAlexa Fluor 568 streptavidin (Invitrogen). In some recordings, SR 95531(Tocris Bioscience) was applied by gravity perfusion at 5 �M in the cir-culating ACSF after being stored frozen as concentrated stock in dH2O.

Statistical analysisStatistical analyses were performed using SPSS 19 and Prism 5. For all behav-ioral experiments, freezing behavior was analyzed using a repeated-measuresANOVA with optogenetic stimulation as the between-subjects factor andtone as the within-subjects factor. Averaged freezing values were also ana-lyzed using a Student’s t test between treatment groups. Activity measure-ments in the open field (time in surround, time in center, distance traveled)were also analyzed using a Student’s t test between treatment groups. Inelectrophysiological experiments, spike number was analyzed using arepeated-measure ANOVA.

ResultsThy1-ChR2-EYFP neurons are a subpopulation ofglutamatergic neurons within the BLATo characterize this population of neurons, we used Thy1-YFPtransgenic mice that have been described previously (Feng et

al., 2000) and a novel transgenic mouse strain, Thy1-ChR2-EYFP (B6.Cg-Tg(Thy1-COP4/EYFP)18Gfng/J), that expresseschannelrhodopsin-2 (ChR2) fused to enhanced YFP also un-der the control of the Thy1 promoter (Arenkiel et al., 2007). Theexpression patterns of YFP and ChR2 driven by the Thy1 pro-moter in both strains are extremely similar within the BLA (Fig.1A,B). YFP and channelrhodopsin-2 expression indicate strongtransgene expression specifically within the BLA, and not withinthe lateral or central amygdala (Fig. 1A,B). To further character-ize this Thy-1-ChR2-expressing subpopulation, we performed insitu hybridization of the Thy1-ChR2-YFP mice, finding that,whereas all Thy1-ChR2-expressing neurons also express vGlut1, amarker of excitatory neurons, only a subset of the vGlut1-expressing neurons express ChR2 (Fig. 1C–H). Further, we usedimmunocytochemistry to analyze the colocalization of anotherexcitatory neuronal marker, CaMKII�, as well as two inhibitorymarkers, parvalbumin and calbindin, with the Thy1-expressingcells. We found that, similar to vGlut1, Thy1-YFP-expressingneurons represent a subset of CaMKII� cells, but do not overlapwith calbindin or parvalbumin populations (Fig. 1I–Q). Overall,these immunofluorescence studies revealed significant overlap inthe localization of Thy1-YFP with a marker for BLA principalneurons, CaMKII�, in the somatodendritic compartment of a largenumber of neurons scattered throughout the BLA. Significantly,there was no overlap with either of the markers for BLA interneu-rons, parvalbumin nor calbindin. These data suggest that the Thy1-YFP and Thy1-ChR2-EYFP neurons represent a distinctsubpopulation of amygdala excitatory, principal neurons. Finally,radiolabeled in situ hybridization of brains from Thy1-ChR2-EYFPmice supported previously reported expression patterns (Feng et al.,2000; Porrero et al., 2010), with particularly strong expression ofChR2 in the BLA (Fig. 2A). Endogenous EYFP expression in Thy1-ChR2-EYFP mice indicates strong transgene expression specificallywithin the BLA, but not LA or central amygdala (CeA) (Fig. 2B).

Electrophysiological correlates of optogenetic activation ofThy1-expressing subpopulation of BLA neuronsActivity of CeM neurons has been linked to increased condi-tioned freezing (Haubensak et al., 2010); thus, we first examinedwhether Thy1-ChR2-YFP-expressing neurons within the BLAcould inhibit CeM neurons.

Current-clamp recordings from BLA Thy1-ChR2-EYFP neu-rons demonstrated high-fidelity spiking in response to 20 Hz laserlight pulses (488 nm, 15 ms, 0.6–1.2 mW), a frequency commonlyused for optogenetic behavioral activation (Fig. 3A; Johansen et al.,2010; Tye et al., 2011; Yizhar et al., 2011). Next, neurons in the CeMwere patch-clamped to examine postsynaptic responses to photoac-tivation of Thy1-ChR2-EYFP-expressing neurons in the BLA (15ms, 0.2 Hz), as well as to bipolar electrical stimulation of the LA (150�s, 0.2 Hz; Fig. 3B). Photoactivation of BLA Thy1-ChR2-EYFP-expressing neurons evoked a postsynaptic potential (PSP) in everyrecorded neuron in the CeM (n � 12). Seven of 12 (58%) CeMneurons exhibited a biphasic response in which an EPSP wasshunted by an IPSP (Fig. 3C). The remaining CeM neurons re-sponded with a monophasic EPSP. Notably, in those CeM neuronsshowing a biphasic response, the IPSP was completely abolished inall neurons tested by bath application of the GABAA receptorantagonist, SR95531 (5 �M; Fig. 3D), and repeated (20 Hz)photoactivation of the BLA elicited a sustained inhibitory re-sponse that abolished spiking in CeM neurons depolarized toaction potential threshold (F(2,4) � 9.89, p � 0.05; Fig. 3E).

Electrical stimulation of the LA evoked a monophasic EPSP ineight of 12 (67%) CeM neurons (Fig. 3C). Notably, in the seven

10398 • J. Neurosci., June 19, 2013 • 33(25):10396 –10404 Jasnow et al. • Cell-Type Specific Inhibition of Fear

CeM neurons that showed a biphasicEPSP/IPSP response to BLA photoactiva-tion, electrical stimulation of the LAevoked either a monophasic EPSP (4 neu-rons) or no response (3 neurons). In thefour CeM neurons that received conver-gent synaptic input from the LA and BLA,photoactivation of the BLA shunted ac-tion potentials driven by electrical stimu-lation of the LA at both low (0.05 Hz; Fig.3C) and high (20 Hz) stimulation fre-quencies (Fig. 3F). Together, these datasuggest that this subpopulation of BLApyramidal neurons inhibits a popula-tion of CeM neurons and can shunt di-rect excitatory drive from the LA.

Activation of Thy1-expressing neuronsinhibits fear consolidationTo examine the behavioral effects of BLAThy1-ChR2-EYFP neuronal activation,mice were implanted with guide cannulaejust dorsal to the BLA to enable light acti-vation of the Thy1-ChR2-EYFP neurons.After recovery and habituation to han-dling, mice underwent fear conditioning.Fear conditioning involved five tone–shock pairings that coterminated with adiscrete 2 s optical activation of Thy1-ChR2-EYFP neurons within the BLA (Fig.4A). We used the same 20 Hz activationfrequency as used for the above slice stud-ies and which had been previously re-ported to be successful in other brainregions when combined with behavior(Arenkiel et al., 2007; Carter et al., 2009;Ciocchi et al., 2010; Gunaydin et al., 2010;Johansen et al., 2010; Zhang et al., 2010).Notably, BLA optical activation was onlypresent during these brief 2 s pulses dur-ing CS–US training period. There was noeffect of optical activation on fear expres-sion during training (Fig. 4B,C).

Twenty-four hours after fear condi-tioning, mice were tested in a novel con-text for freezing responses to the CS. Micethat had previously received CS–US train-ing coterminating with laser photoactiva-tion displayed significantly less freezing inresponse to CS presentation comparedwith mice receiving sham stimulation(F(1,12) � 17.84, p � 0.001; t12 � 4.224,p � 0.001; Fig. 4B,C). These data suggestthat photoactivation of BLA Thy1-ChR2-EYFP neurons reduced long-term fearmemory consolidation. These results areconsistent with the hypotheses that opto-genetic activation of the Thy1 subpopula-tion may decrease general fear/anxietyresponses, interfere with amygdala func-tioning, or specifically inhibit condi-tioned fear. We sought to separate thesepossibilities with subsequent experi-

Figure 1. Molecular characterization of Thy1-expressing pyramidal neurons within basolateral amygdala transgene expression drivenby the Thy1 promoter in two lines of mice (Thy1-YFP, Thy1-ChR2-EYFP). Extremely similar expression patterns of the two transgenes drivenby the Thy1 promoter are observed in the BLA between the two lines. Although the expression pattern is the same, more cells appearlabeled in A because of the thickness of the section (45 �m in A vs 16 �m in B). A, A coronal section of a Thy1-YFP mouse, double-labeledwithDAPI(blue)stainingallcellsandYFPlabelingtheThy1subpopulation, inwhichtheThy1-YFPpyramidalneuronsareonly locatedintheBLA, and not the LA or CeA regions. B, A coronal section of a Thy1-ChR2-EYFP mouse identifying ChR2 transgene expression (green) usingfluorescent in situ hybridization in which the Thy1-ChR2 neurons are only located in the BLA, and not the LA or CeA regions. Dual in situhybridization for ChR2 and Vglut1 demonstrating that ChR2 neurons in the BLA are glutamatergic. C–E, Coronal sections (10�) of aThy1-ChR2-EYFP mouse labeled with VGlut1 mRNA (green, C), ChR2 mRNA (red, D), and merged (E). F–H, Coronal sections (20�) of aThy1-ChR2-EYFP mouse labeled with VGlut1 mRNA (green, F ), ChR2 mRNA (red, G), and merged (H ). All Vglut1 neurons do not expressChR2; however, virtually all ChR2 neurons express Vglut1 within the BLA. I–Q, Dual immunocytochemistry characterization of the Thy1-EYFP neuronal population in the Thy1-EYFP mouse line (green) and its colocalization with a subset of CaMKII� neurons (I–K ), but lack ofcolocalization with the inhibitory calbindin (L–N ) or parvalbumin (O–Q) populations.

Jasnow et al. • Cell-Type Specific Inhibition of Fear J. Neurosci., June 19, 2013 • 33(25):10396 –10404 • 10399

ments. To determine whether tissue heat-ing/damage was the cause of the blockadeof fear consolidation, we replicated theabove experiments using Thy1-YFP mice,which lack ChR2 expression. Here weshow that optogenetic stimulation in micelacking ChR2 does not inhibit condi-tioned fear (Fig. 4D). Thus, the above ef-fect was unlikely due to tissue heating ordamage. Previous data suggested that opto-genetic stimulation of the lateral amygdalacould serve as an unconditioned stimulus(Johansen et al., 2010). Thus, to determinewhether optogenetic stimulation of Thy1-ChR2-expressing neurons within the BLAcould serve as an unconditioned stimulus,we paired tone CS presentations with opto-genetic stimulation of the BLA and testedthe mice for freezing 24 h later. Freezing re-sponses during testing were negligible, sug-gesting that optogenetic stimulation ofThy1-ChR2-expressing neurons withinthe BLA is unable to support fear learning inthe absence of true unconditioned stimulus(Fig. 4E).

Next, to determine whether the inhibi-tion of freezing was due to a general effecton anxiety-like or unconditioned fear re-sponses, we observed behavior and loco-motion in an open-field paradigm duringoptogenetic activation of Thy1-ChR2-EYFP cells in the BLA. Mice were tested inan open field in the absence or presence ofthe same 2 s optical stimulation parame-ters used during fear conditioning (Fig.4F). Optogenetic stimulation of the BLAdid not cause any significant alteration inlocomotion or anxiety-like behavior, asmeasured by total distance traveled (t5 �0.19, p � 0.05), time spent in the center ofthe open field (t5 � 0.69, p � 0.05), ortime spent in the surround (t5 � 0.62, p � 0.05; Fig. 4F). Inaddition, there were no time-dependent differences in theseopen-field measures over the first 5 min period in the absence ofoptogenetic stimulation (ps � 0.05; Fig. 4F), suggesting that de-creased neophobia did not mask an apparent effect of optoge-netic stimulation on anxiety. Combined, these data suggest thatour discrete optical stimulation parameters, when coterminatedwith tone and shock, likely have a specific inhibitory effect on theconsolidation of long-term fear memory. Moreover, these datasuggest that Thy1-ChR2-EYFP-expressing neurons may be partof a specific neural circuit inhibiting fear.

Activation of Thy1-expressing neurons enhancesconsolidation of fear extinctionIf Thy1-ChR2-expressing neurons are part of a specific neuralcircuit inhibiting fear, then they should also enhance extinctionlearning. In contrast, if this activation were leading to a gen-eral disruption of amygdala processing, we would expect to seea disruption of extinction, as amygdala plasticity is required forextinction learning. To test this hypothesis, mice were trained ina cued fear conditioning paradigm, then 24 h later were extinctiontrained, followed by testing for extinction retention. Mice were fear

conditioned with tone–shock pairings alone and then received opti-cal stimulation during extinction training. Fifteen presentations ofthe tone CS coterminated with a 2 s optical stimulation of the BLAduring extinction training (Fig. 5A), using the same optical stimula-tion parameters described above. During extinction training, opticalactivation of Thy1-ChR2 neurons in the BLA had no effect on freez-ing (F(1,8) � 0.019, p � 0.05; t8 � 0.1821, p � 0.05; Fig. 5B,C), againindicating that our discrete activation parameters had no effect onfear expression or freezing behavior alone.

However, when the mice were retested for extinction reten-tion in the absence of optical stimulation, mice previously extinc-tion trained with optical stimulation displayed significantly lessfear compared with mice receiving sham stimulation. This sug-gests that optical stimulation of BLA Thy1-ChR2-EYFP neu-rons enhanced extinction consolidation (F(1,8) � 9.0, p � 0.017;t8 � 2.972, p � 0.017; Fig. 5B,C). Thus, optical stimulationwithin the BLA did not result in a general impairment ofamygdala functioning due to damage or enhanced feedbackinhibition onto interneurons producing reduced learning orplasticity, two potential explanations for the deficit in fearlearning we observed previously. In both cases, we would alsoexpect to see an inhibition of extinction of fear rather than the

Figure 2. Channel rhodopsin and EYFP expression within the basolateral amygdala of the Thy1-ChR2-EYFP mouse line. A, Autoradiog-raphy of an in situ hybridization with a 35-S-labeled riboprobe targeting the channelrhodopsin (ChR2) gene from anterior (left) to moreposterior (right) coronal sections including BLA. B, EYFP labeling of Thy1-ChR2-EYFP cells and fibers from anterior (topr left) to posterior(bottem right) coronal sections spanning the amygdala, demonstrating Thy1-ChR2-EYFP labeling in BLA, but not LA or CeA nuclei of theamygdala. This pattern of ChR2 and EYFP expression recapitulates what has been observed in the Thy1-EYFP line. C, C-fos immunostainingwas performed �1 h after laser light stimulation of Thy1-ChR2-EYFP mice targeting the BLA, following the behavioral studies describedbelow, using similar stimulation paradigms. We found that without stimulation, there was minimal c-fos activation, whereas with stimu-lation, there was increased c-fos staining in BLA and CeL. LaDL, Dorsolateral amygdala.

10400 • J. Neurosci., June 19, 2013 • 33(25):10396 –10404 Jasnow et al. • Cell-Type Specific Inhibition of Fear

enhanced fear extinction we observed (note that a number ofother manipulations such as pharmacological blockade ofamygdala functioning, blockade of NMDA functioning, oramygdala lesions all inhibit both fear learning and extinctionlearning; Myers and Davis, 2006; Johansen et al., 2011). There-fore, these data are consistent with a relatively specific effect ofBLA Thy1-ChR2-EYFP neuronal activation on enhancing in-hibitory/extinction learning.

We next tested whether the enhanced extinction consolida-tion would transfer to a new context. It is generally accepted thatextinction is a form of context-dependent learning, not an era-sure of the original fear memory (Myers et al., 2006). Placingmice back into the original training context typically results ina return of freezing, termed “fear renewal”. Here, we demon-strate that optogenetic-induced enhanced extinction consolidationalso inhibits renewal of fear (F(1,8) � 9.4, p � 0.01; Fig. 5D). Thesedata suggest that the enhanced extinction learning is potentially due

to a strengthening of the CS–no-US associ-ation, rendering context to be less relevantand enabling the transfer of extinctionacross contexts.

One could argue that the blockade ofspontaneous recovery (enhanced ex-tinction retention) and renewal of fearwith optogenetic stimulation of BLAThy1-ChR2-EYFP neurons representserasure or unlearning of the fear memory.To test this possibility, we examined bothgroups of mice in a reinstatement test.Mice that were previously extinctiontrained were subjected to a single unsig-naled footshock in an entirely new con-text. We then assessed tone-evokedfreezing 24 h later (reinstatement test).Here, we demonstrate that the fear mem-ory was fully reinstated following a singleunsignaled footshock (F(1,8) � 0.19, p �0.05; Fig. 5D), consistent with a savings ofthe prior memory trace, indicating thatthe fear memory was not erased. Rather,optogenetic stimulation of Thy1-ChR2-EYFP-expressing BLA neurons appearsto directly enhance fear extinction con-solidation and may also render theextinction memory independent of con-text modulation, which is critical to theprocess of fear renewal. In addition, micecan express fear normally and can be re-trained to an additional cue followingoptogenetic-enhanced extinction, suggest-ing that activation of Thy1-ChR2-EYFP-expressing neurons does not permanentlyalter amygdala function.

DiscussionOur data demonstrate that pairing activa-tion of a specific neuronal populationonly during the CS–US presentation in-hibited the memory for the associationbetween the CS and US, and thus inhib-ited fear consolidation. Conversely, dur-ing extinction training, pairing activationof Thy1-ChR2-expressing neurons dur-ing a very discrete time window during

presentation of the CS alone enhanced the CS–no-US associa-tion, and thus enhanced extinction retention. It is important tonote the associative nature of the pairing of the optogenetic stim-ulation with the CS–US or CS–no-US as compared to previousreports using prolonged stimulation parameters examining fearand anxiety-like behavior (Goshen et al., 2011; Tye et al., 2011).The present data suggest that the BLA contains a subpopulationof glutamatergic neurons that may be specialized for inhibitingfear responses.

We demonstrate that a Thy1-ChR2-EYFP-expressing sub-population of BLA neurons can inhibit CeM neuronal activity.In contrast, electrical stimulation of the LA evoked a purelyexcitatory response in CeM neurons. Significantly, concurrentoptogenetic activation of Thy1-ChR2-EYFP neurons withinthe BLA shunted the excitatory LA driven response and pre-vented CeM neuron activation. Discrete optogenetic activation

Figure 3. Electrophysiological effects within the BLA and CeM of optogenetic activation of BLA-Thy1 neurons. A, A represen-tative current-clamp recording depicts typical, high-fidelity spiking of a BLA-Thy1 pyramidal neuron in response to laser lightactivation (blue bars) at 20 Hz with a pulse width of 15 ms. B, Slice electrophysiology experiments were performed on Thy1-ChR2-EYFP mice with neurons patch clamped in the CeM, a fiber-optic light source targeting the BLA, and a bipolar stimulating electrodeplaced in the LA. A diagram of the experimental setup is overlaid on a photomicrograph depicting YFP expression (green) in Thy1neurons of the BLA and a recorded, biocytin-filled neuron in the CeM (red). C, A representative current-clamp recording from a CeMneuron depicts a biphasic response (left) to optogenetic BLA-Thy1 activation (blue bar) and a shunt of an action potential elicitedby LA stimulation (stim). Responses without action potentials were averaged for five consecutive sweeps (n � 12). D, A represen-tative, biphasic response in a CeM neuron, held at a membrane potential of �55 mV with direct current injection, to lightactivation of BLA-Thy1 neurons (left). The response became purely depolarizing following bath application of 5 �M SR95531 andwas able to elicit an action potential (right) (n�3). E, A representative current-clamp recording of a free-firing CeM neuron, depolarized toaction potential threshold with direct current, depicts blockade of ongoing spiking by a 20 Hz, 1 s train of optogenetic BLA-Thy1 activation(blue bars) (n � 12). F, Current-clamp recordings of a CeM neuron, depolarized near threshold, depict 20 Hz optogenetic activation (bluebars) of BLA-Thy1 neurons inhibiting the recorded cell (top) and shunting LA-stimulation-induced excitation (bottom), which itself evokedaction potentials (middle). Stimulation artifacts were cropped at �25 and �50 mV for clarity.

Jasnow et al. • Cell-Type Specific Inhibition of Fear J. Neurosci., June 19, 2013 • 33(25):10396 –10404 • 10401

of Thy1-ChR2-EYFP-expressing neuronsin the BLA during cued fear conditioninginhibits fear consolidation, whereasactivation during suboptimal fear ex-tinction training enhances the con-solidation of the extinguished fear re-sponse. The observed enhancement ofextinction learning was sufficient to allowtransfer of the extinguished response tothe training context in a fear renewal test.This did not, however, disrupt the initialmemory trace, because the fear responsewas fully reinstated with a single unsig-naled footshock. It is unlikely that the ob-served extinction enhancement was dueto a blockade of reconsolidation becauseof the duration of exposure to the CS inthe present study. Several recent studiesdemonstrate that one of the key factors indistinguishing reconsolidation processesfrom extinction processes is the duration ofCS exposure. In the present study, micewere exposed to 15 CS tones over a periodof 22 min, which is beyond what hasbeen reported to initiate reconsolida-tion, but does initiate extinction (Pe-dreira and Maldonado, 2003; Suzuki etal., 2004; Bredy and Barad, 2008; Perez-Cuesta and Maldonado, 2009; Kirtleyand Thomas, 2010; Inda et al., 2011; dela Fuente et al., 2011). Thus, despite thelong-held role of excitatory neurotrans-mission within the BLA in fear, the cur-rent study demonstrates that activationof a population of glutamatergic neu-rons within the BLA specifically inhibitsfear, apparently without disrupting theoriginal memory trace.

Importantly, activation of Thy1-ChR2-EYFP-expressing neurons during the open-field test did not reduce anxiety-likebehavior or alter locomotion. Althoughpossible, this lack of effect on anxiety wasunlikely due to a lack of neophobia duringthe second 5 min period of testing, asanxiety-like behavior did not vary as a func-tion of time during the unstimulated time ofthe open-field test (Fig. 4F). In addition, op-togenetic stimulation was unable to supportfear learning in the absence of an uncon-ditioned stimulus (Fig. 4E), indicatingbehavioral specificity of our stimulationparameters. In another study, optical activa-tion of the lateral amygdala paired with pre-sentation of a conditioned stimulus duringtraining resulted in an increase in freezing during testing 24 h later(Johansen et al., 2010), suggesting that activation of LA pyramidalneurons is sufficient to produce long-term fear memory when asso-ciated with a CS in the absence of a US. These data further supportour electrophysiological results of electrical stimulation of the LAactivating CeM neurons, which could be shunted by optogeneticstimulation of BLA Thy1-ChR2-expressing neurons (Fig. 2). As ac-tivity of CeM neurons has been linked to increased conditioned

freezing (Haubensak et al., 2010), our electrophysiological and be-havioral data fit nicely with the role of this nucleus regulating fearresponses.

Our findings are consistent with an earlier report (Herry et al.,2008) suggesting that the previously identified extinction neu-rons represent a subset of a larger population of neurons withinthe BLA that are more generally involved in inhibiting fear mem-ory. Interestingly, the slice physiology data suggest that some

Figure 4. Discrete optogenetic activation of BLA Thy1 neurons inhibits fear memory formation without locomotor or anxiety-like effects. A, Training paradigm. Mice were trained with five trials of tones (30 s, 6 kHz) which coterminated with a 1 s footshock(0.5 mA). Thy1-ChR2-EYFP mice received either sham stimulation or the 30 s tone also coterminated with a 2 s (20 Hz, 15 ms) bluelight targeting BLA. Left, EYFP expression in BLA. Right, Anticipated light cone with fiber optic cable. B, Freezing during trainingand 24 h later when mice were tested (averaged blocks of 2 trials) in the absence of optogenetic activation and in the presence ofthe conditioned tone cues. Mice that had been trained in conjunction with optogenetic activation (n � 8) show minimal expres-sion of fear (n�7 controls), C, Representation of data from A as a bar graph. Left, Average freezing responses to the last three tonesduring training and in the presence or absence of optogenetic stimulation. Right, Average freezing responses from A during testingin the absence of optogenetic stimulation. Mice that had been trained in conjunction with optogenetic activation show minimalexpression of fear. D, Optogenetic stimulation in Thy1-EYFP mice lacking ChR2 expression has no effect on conditioned fear. Left,Average freezing responses to the last three tones during training and in the presence (n � 8) and absence (n � 9) of optogeneticstimulation. Right, Average freezing responses during testing in the absence of optogenetic stimulation. E, Optogenetic stimula-tion during training in Thy1-ChR2-EYFP mice in the absence of a US does not support fear learning (n � 7), thus optogeneticstimulation of Thy1-expressing neurons of the BLA alone cannot serve as a US. Bar graph represents average freezing responses to10 tone presentations. F, Mice received the same stimulation procedure (2 s, 20 Hz, 15 ms blue light) during a standard open-fieldtesting procedure. Left, Locomotion in the open field [total distance traveled (Tr.)] in control (n � 6) versus stimulated groups(n � 6) of Thy1-ChR2-EYFP mice. Center, Anxiety-like behavior (time in center of open-field) in control versus stimulated groups.Right, Time in surround within open field in control versus stimulated groups. Additionally, there were no time-dependent changesin neophobia during the behavioral test that would mask a lack of effect of optogenetic stimulation during the second 5 min period.White and gray bars represent the same measures as above during the first minute and fourth minute, respectively, within theinitial unstimulated testing period. Error bars represent mean SEM. *p � 0.05.

10402 • J. Neurosci., June 19, 2013 • 33(25):10396 –10404 Jasnow et al. • Cell-Type Specific Inhibition of Fear

neurons in the CeM are inhibited while others are excited byoptogenetic photoactivation of BLA Thy1-ChR2-EYFP neurons.In these studies of auditory fear, Thy1-ChR2-EYFP neuronal ac-tivation within the BLA appears to selectively support inhibitionof fear learning and enhancement of fear extinction. One possiblemechanism by which Thy1-ChR2-EYFP neurons inhibit fearmay be via indirect inhibition of neurons in the CeM, as weshowed for a number of CeM neurons upon optogenetic activa-tion of these BLA neurons. Indeed, in a previous report, opticalactivation of BLA fibers within the lateral CeA inhibited anxiety-like behavior, whereas inhibition of the same fibers enhancedanxiety-like behavior (Tye et al., 2011). Our current data supporta similar mechanism for fear inhibition. Here, we observed feed-forward inhibition of CeM neurons via optical stimulation ofThy1-ChR2-expressing BLA neurons (Fig. 3). It is likely that thatthe Thy1-ChR2-EYFP neurons examined in the present studyrepresent a similar population of neurons driving CeA responsesinvolved in anxiety-like behavior, although this was not defini-tively determined. It is also possible that Thy1-ChR2-EYFP-expressing neurons represent a microcircuit regulating directlocal responses within the BLA, but are synaptically connected toneurons that project to the CeA.

In the Herry and colleagues’ (2008) study, extinction neuronsrepresented 14% of total recorded units, which is a much smallerpopulation than those activated here. In addition, extinctionneurons are preferentially synaptically connected to the mPFC,placing them in an ideal position to regulate fear extinction.Thy1-ChR2-EYFP-expressing neurons make up a sizable portionof large spiny glutamatergic neurons in the BLA, as identifiedwith in situ hybridization and immunohistochemistry in Thy1-YFP transgenic mice. Currently, it is unknown whether Thy1-YFP-expressing neurons of the BLA receive afferent inputs fromthe mPFC, or how they may be different from other glutamater-gic neurons within the BLA. However, Thy1-YFP-expressingBLA neurons also send efferent fibers along the stria terminalis,terminating primarily in the nucleus accumbens (NAc) and cau-date nucleus, as well as the bed nucleus of the stria terminalis(Porrero et al., 2010). The nucleus accumbens has previouslybeen implicated in fear extinction (Pezze and Feldon, 2004;Holtzman-Assif et al., 2010; Muschamp et al., 2011), suggestingan essential role of this BLA–NAc pathway in fear inhibition/extinction. Overall, the circuit connectivity between mPFC, NAc,and BLA may act in concert with the CS-responding Thy1-YFPneurons to drive extinction learning in contrast to supportingcontinued fear expression.

Notably, there is a wealth of prior data showing that simplyactivating or inactivating the amygdala as a whole does not pro-duce the patterns of functional responses that we have observed.Prior data demonstrate that general inhibition of the BLAthrough GABA agonists or NMDA antagonists block the acqui-sition/consolidation of fear learning (Wilensky et al., 1999) aswell as block the extinction of fear (Falls et al., 1992). There is alsoevidence that general pharmacological enhancement of NMDA

D

Renewal Reinstatement

*

* *

*

B Extinction Ext. Retention Blocks of 2

*

%Fr

eezi

ng

%Fr

eezi

ng

%Fr

eezi

ng

%Fr

eezi

ng

Extinction Ext. Retention

C

B

A

*

Tone Tone/Light

30 sec, 6 KHz Tone X 5

Training

30 sec, 6 KHz Tone X 15Testing

24 hr= Shock= Laser

Figure 5. Discrete optogenetic activation of Thy1 neurons within BLA enhances extinctionlearning. A, Top, Fear training paradigm. All cannulated Thy1-ChR2-EYFP mice were trainedwith five trials of tones (30 s, 6 kHz), which coterminated with a 0.5 s footshock (0.5 mA).Bottom, Twenty-four hours later, in a separate context, all mice received 15 trials of tones (30 s,6 kHz) coterminating with either sham stimulation or with a 2 s, BLA-targeted, blue light (20 Hz,15 ms). B, No difference between sham versus optogenetically activated groups was seenduring extinction training (B, Baseline freezing during extinction before CS presentation). How-ever, 24 h later, mice that received the blue light during extinction (n � 5) showed much lowerfreezing to the cues compared with controls (n � 5), consistent with dense extinction (Ext.)retention. C, Representation of data from B as a bar graph. Left, Average freezing responsesduring extinction to 15 tone presentations and in the presence or absence of optogenetic stim-ulation. Right, Average freezing responses from B during extinction retention testing in the

4

absence of optogenetic stimulation. Mice that had been extinction trained in conjunction withoptogenetic activation showed enhanced extinction retention. D, Twenty-four hours later, micewere fear tested in the initial conditioned context. Those without optogenetic stimulation (n �5) showed renewal of fear, whereas those with optogenetic stimulation (n � 5) had minimalfreezing to the tone. Following footshock reinstatement in a new context, all mice showedsimilar levels of tone-activated freezing behavior. Error bars represent mean SEM.*p � 0.05.

Jasnow et al. • Cell-Type Specific Inhibition of Fear J. Neurosci., June 19, 2013 • 33(25):10396 –10404 • 10403

function in the BLA enhances fear acquisition (Maren et al.,1996) as well as fear extinction (Walker et al., 2002). Thus, ourresults are most parsimoniously interpreted as the specific acti-vation of a functional subpopulation of pyramidal neurons.

In conclusion, these data suggest that activation of specific,genetically marked subpopulations of principal neurons withinthe BLA may selectively inhibit fear. It should be noted, however,that the use of transgenics in the current study do not allow us torule out the possibility that stimulation of fibers passing throughthe BLA resulted in some of the observed electrophysiological orbehavioral effects. Further work remains to be done to more fullycharacterize the circuit within which Thy1-ChR2-EYFP neuronsinhibit fear and to understand whether this property is shared byall of the Thy1-YFP subpopulation or just those whose activationleads to CeM inhibition. Identification of molecular markersof neurons that selectively inhibit fear or enhance extinctionwould be particularly useful for enhancing our understandingof fear processes. Additionally, cell-type-specific amplification ofmRNA may allow the identification of specific receptor targetsthat are preferentially expressed on this subset of cells, potentiallyleading to novel targets for disorders of fear regulation.

ReferencesArenkiel BR, Peca J, Davison IG, Feliciano C, Deisseroth K, Augustine GJ,

Ehlers MD, Feng G (2007) In vivo light-induced activation of neuralcircuitry in transgenic mice expressing channelrhodopsin-2. Neuron 54:205–218. CrossRef Medline

American Psychiatric Association (2000) Diagnostic and statistical manualof mental disorders, 4 ed. Washington, DC: American PsychiatricAssociation.

Bouton ME, Westbrook RF, Corcoran KA, Maren S (2006) Contextual andtemporal modulation of extinction: behavioral and biological mecha-nisms. Biol Psychiatry 60:352–360. CrossRef Medline

Bredy TW, Barad M (2008) The histone deacetylase inhibitor valproic acidenhances acquisition, extinction, and reconsolidation of conditionedfear. Learn Mem 15:39 – 45. CrossRef Medline

Carter ME, Adamantidis A, Ohtsu H, Deisseroth K, de Lecea L (2009) Sleephomeostasis modulates hypocretin-mediated sleep-to-wake transitions.J Neurosci 29:10939 –10949. CrossRef Medline

Ciocchi S, Herry C, Grenier F, Wolff SB, Letzkus JJ, Vlachos I, Ehrlich I,Sprengel R, Deisseroth K, Stadler MB, Muller C, Luthi A (2010) Encod-ing of conditioned fear in central amygdala inhibitory circuits. Nature468:277–282. CrossRef Medline

Davis M (2000) The role of the amygdala in conditioned and unconditionedfear and anxiety. In: The amygdala (Aggleton J 2nd, ed.), pp 213–287.Oxford: Oxford UP.

Falls WA, Miserendino MJ, Davis M (1992) Extinction of fear-potentiatedstartle: blockade by infusion of an NMDA antagonist into the amygdala.J Neurosci 12:854 – 863. Medline

Feng G, Mellor RH, Bernstein M, Keller-Peck C, Nguyen QT, Wallace M,Nerbonne JM, Lichtman JW, Sanes JR (2000) Imaging neuronal subsetsin transgenic mice expressing multiple spectral variants of GFP. Neuron28:41–51. CrossRef Medline

Goshen I, Brodsky M, Prakash R, Wallace J, Gradinaru V, Ramakrishnan C,Deisseroth K (2011) Dynamics of retrieval strategies for remote memo-ries. Cell 147:678 – 689. CrossRef Medline

Gunaydin LA, Yizhar O, Berndt A, Sohal VS, Deisseroth K, Hegemann P(2010) Ultrafast optogenetic control. Nat Neurosci 13:387–392.CrossRef Medline

Haubensak W, Kunwar PS, Cai H, Ciocchi S, Wall NR, Ponnusamy R, Biag J,Dong HW, Deisseroth K, Callaway EM, Fanselow MS, Luthi A, AndersonDJ (2010) Genetic dissection of an amygdala microcircuit that gatesconditioned fear. Nature 468:270 –276. CrossRef Medline

Herry C, Ciocchi S, Senn V, Demmou L, Muller C, Luthi A (2008) Switchingon and off fear by distinct neuronal circuits. Nature 454:600 – 606.CrossRef Medline

Herry C, Ferraguti F, Singewald N, Letzkus JJ, Ehrlich I, Luthi A (2010)Neuronal circuits of fear extinction. Eur J Neurosci 31:599 – 612. CrossRefMedline

Holtzman-Assif O, Laurent V, Westbrook RF (2010) Blockade of dopamineactivity in the nucleus accumbens impairs learning extinction of condi-tioned fear. Learn Mem 17:71–75. CrossRef Medline

Inda MC, Muravieva EV, Alberini CM (2011) Memory retrieval and thepassage of time: from reconsolidation and strengthening to extinction.J Neurosci 31:1635–1643. CrossRef Medline

Johansen JP, Hamanaka H, Monfils MH, Behnia R, Deisseroth K, BlairHT, LeDoux JE (2010) Optical activation of lateral amygdala pyra-midal cells instructs associative fear learning. Proc Natl Acad Sci U S A107:12692–12697. CrossRef Medline

Johansen JP, Cain CK, Ostroff LE, LeDoux JE (2011) Molecular mecha-nisms of fear learning and memory. Cell 147:509 –524. CrossRef Medline

Kessler RC, Chiu WT, Demler O, Merikangas KR, Walters EE (2005)Prevalence, severity, and comorbidity of 12-month DSM-IV disordersin the National Comorbidity Survey Replication. Arch Gen Psychiatry62:617– 627. CrossRef Medline

Kirtley A, Thomas KL (2010) The exclusive induction of extinction is gatedby BDNF. Learn Mem 17:612– 619. CrossRef Medline

de la Fuente V, Freudenthal R, Romano A (2011) Reconsolidation or extinc-tion: transcription factor switch in the determination of memory courseafter retrieval. J Neurosci 31:5562–5573. CrossRef Medline

LeDoux JE (2000) Emotion circuits in the brain. Annu Rev Neurosci 23:155–184. CrossRef Medline

Maren S, Fanselow MS (1996) The amygdala and fear conditioning: has thenut been cracked? Neuron 16:237–240. CrossRef Medline

Maren S, Aharonov G, Stote DL, Fanselow MS (1996) N-methyl-D-aspartate receptors in the basolateral amygdala are required for both ac-quisition and expression of conditional fear in rats. Behav Neurosci 110:1365–1374. CrossRef Medline

McDonald A, Muller J, Mascagni F (2002) GABAergic innervation of alphatype II calcium/calmodulin-dependent protein kinase immunoreactivepyramidal neurons in the rat basolateral amygdala. J Comp Neurol 446:199 –218. CrossRef Medline

Muschamp JW, Van’t Veer A, Parsegian A, Gallo MS, Chen M, Neve RL,Meloni EG, Carlezon WA Jr (2011) Activation of CREB in the nucleusaccumbens shell produces anhedonia and resistance to extinction of fearin rats. J Neurosci 31:3095–3103. CrossRef Medline

Myers KM, Davis M (2006) Mechanisms of fear extinction. Mol Psychiatry12:120 –150. CrossRef Medline

Myers KM, Ressler KJ, Davis M (2006) Different mechanisms of fear extinctiondependent on length of time since fear acquisition. Learn Mem 13:216–223.CrossRef Medline

Paxinos G, Franklin KBJ (2004) The mouse brain in stereotaxic coordinates.San Diego: Academic.

Pedreira ME, Maldonado H (2003) Protein synthesis subserves reconsolida-tion or extinction depending on reminder duration. Neuron 38:863– 869.CrossRef Medline

Perez-Cuesta LM, Maldonado H (2009) Memory reconsolidation and ex-tinction in the crab: mutual exclusion or coexistence? Learn Mem 16:714 –721. CrossRef Medline

Pezze MA, Feldon J (2004) Mesolimbic dopaminergic pathways in fear con-ditioning. Prog Neurobiol 74:301–320. CrossRef Medline

Porrero C, Rubio-Garrido P, Avendano C, Clasca F (2010) Mapping of flu-orescent protein-expressing neurons and axon pathways in adult anddeveloping Thy1-eYFP-H transgenic mice. Brain Res 1345:59 –72.CrossRef Medline

Sugino K, Hempel CM, Miller MN, Hattox AM, Shapiro P, Wu C, Huang ZJ,Nelson SB (2006) Molecular taxonomy of major neuronal classes in theadult mouse forebrain. Nat Neurosci 9:99 –107. CrossRef Medline

Suzuki A, Josselyn SA, Frankland PW, Masushige S, Silva AJ, Kida S (2004)Memory reconsolidation and extinction have distinct temporal and bio-chemical signatures. J Neurosci 24:4787– 4795. CrossRef Medline

Tye KM, Prakash R, Kim SY, Fenno LE, Grosenick L, Zarabi H, ThompsonKR, Gradinaru V, Ramakrishnan C, Deisseroth K (2011) Amygdala cir-cuitry mediating reversible and bidirectional control of anxiety. Nature471:358 –362. CrossRef Medline

Walker DL, Ressler KJ, Lu KT, Davis M (2002) Facilitation of conditionedfear extinction by systemic administration or intra-amygdala infusions ofD-cycloserine as assessed with fear-potentiated startle in rats. J Neurosci22:2343–2351. Medline

Wilensky AE, Schafe GE, LeDoux JE (1999) Functional inactivation of the

10404 • J. Neurosci., June 19, 2013 • 33(25):10396 –10404 Jasnow et al. • Cell-Type Specific Inhibition of Fear

amygdala before but not after auditory fear conditioning prevents mem-ory formation. J Neurosci 19:48RC.

Yizhar O, Fenno LE, Davidson TJ, Mogri M, Deisseroth K (2011) Optoge-netics in neural systems. Neuron 71:9 –34. CrossRef Medline

Zhang F, Gradinaru V, Adamantidis AR, Durand R, Airan RD, de Lecea L,Deisseroth K (2010) Optogenetic interrogation of neural circuits: tech-nology for probing mammalian brain structures. Nat Protoc 5:439 – 456.CrossRef Medline

Jasnow et al. • Cell-Type Specific Inhibition of Fear J. Neurosci., June 19, 2013 • 33(25):10396 –10404 • 10404a